KNX bus did not respond in time (although everything works)

tominvasio · July 2, 2025, 7:28am

Hey there!

I see there are several similar posts on the forum. However I can’t fully identify with any of them.

My household is (also) based on KNX. Everything works flawlessly (which does not seem to be the case in other reports). States in HA get updated, whenever some sensor changes it’s state. But the logbook concerns me. I see messages like this coming in every few seconds:

2025-07-02 08:16:32.285 WARNING (MainThread) [xknx.log] Error: KNX bus did not respond in time (2.0 secs) to GroupValueRead request for: 8/0/1
2025-07-02 08:16:32.285 WARNING (MainThread) [xknx.log] Could not sync group address '8/0/1' (TF AZ T2 - State)
2025-07-02 08:17:21.373 WARNING (MainThread) [xknx.log] Error: KNX bus did not respond in time (2.0 secs) to GroupValueRead request for: 4/0/1
2025-07-02 08:17:21.373 WARNING (MainThread) [xknx.log] Could not sync group address '4/0/1' (TF WK T2 - State)
2025-07-02 08:17:32.866 WARNING (MainThread) [xknx.log] Error: KNX bus did not respond in time (2.0 secs) to GroupValueRead request for: 0/0/2
2025-07-02 08:17:32.866 WARNING (MainThread) [xknx.log] Could not sync group address '0/0/2' (TF Küche Tür T2 - State)
2025-07-02 08:17:33.330 WARNING (MainThread) [xknx.log] Error: KNX bus did not respond in time (2.0 secs) to GroupValueRead request for: 0/0/1
2025-07-02 08:17:33.330 WARNING (MainThread) [xknx.log] Could not sync group address '0/0/1' (TF Küche Tür T1 - State)
2025-07-02 08:25:30.991 WARNING (MainThread) [xknx.log] Error: KNX bus did not respond in time (2.0 secs) to GroupValueRead request for: 14/2/22
2025-07-02 08:25:30.991 WARNING (MainThread) [xknx.log] Could not sync group address '14/2/22' (Kontaktschalter SZ Fenster links - State)

This goes on all day and seems to occur with all configured addresses.

My hardware and ETS setup
I have a KNX Interface SCN-IP000.03 from MDT with firmware 3.0.8 (3.1.7 would be available) installed. The addressing is set up as follows:

Individual Address: 1. 1. 35
Tunneling Addresses:
- 15.15.241 → UDP: 192.168.178.128 (Raspberry Pi)
- 15.15.242 → not in use
- 15.15.243 → not in use
- 15.15.244 → TCP: 192.168.178.120 (Home Assistant)

In ETS, I can’t find any relevant settings that might influence the situation (but that doesn’t mean there aren’t any ).

My Home Assistant setup
Home Assistant OS with Core 2025.6.3 inside a Proxmox machine.

KNX interface configured as tunnel
- Interface: 1.1.35 - MDT KNX TP IP @ 192.168.178.5:3671 TCP
- Endpoint: 15.15.244 (as shown also on the addresses list above)
KNX communication settings
- Rate limit: 10 (I think I reduced that sometime in the past)
- Telegram history limit 200

I have another KNX IP interface available with a Loxone controller. However, when I swith on that, I see the same logbook entries. So no improvement.

I’ve done some minutes of debug logging. There may be there’s some relevant information in there, but it’s too complex for me to understand…

I would be grateful for any help in improving my environment.

Thomas

DeejayT · July 2, 2025, 7:34am

I have the same error. Over and over again, “could not sync group address x/x/x.” This happens with all addresses. I can’t identify a pattern.

tominvasio · July 2, 2025, 7:47am

Good to know I’m not alone

farmio · July 2, 2025, 8:02am

Hi !
I’d recommend to set appropriate addresses for your tunnel endpoints - something free on the line it actually is connected to. J

Don’t.

Other than that, have a look at https://www.home-assistant.io/integrations/knx/#group-address-can-not-be-read

tominvasio · July 2, 2025, 8:41am

Thanks for participating, @farmio!

Can you help me out? I’m not so deep into KNX… What would be appropriate addresses and which need to be changed?

I’ve set it now to 0 to disable the rate limit.

Didn’t stumble over that one yet Seems to hold some valuable information. For my understanding, may the missing Read flag responsible for the log entries?

farmio · July 2, 2025, 8:46am

If your device is at 1.1 an appropriate address would be 1.1.240 (1.1.x). Not 15.15.x

Yes, but it has to be set on the correct group object. Usually those have proper defaults…

tominvasio · July 2, 2025, 9:28am

If I get you right, you are referring to these settings:

Individual Address: 1.1.35
Tunneling Addresses:
- 15.15.241 → UDP: 192.168.178.128 (Raspberry Pi)
- 15.15.242 → not in use
- 15.15.243 → not in use
- 15.15.244 → TCP: 192.168.178.120 (Home Assistant)

Which would mean to set the tunneling addresses from 15.15.xxx to (available) 1.1.xxx.

Now, as example, there’s a wall pushbutton and an integrated temperature sensor. The temperature sensor is the only object, where the Read Flag is set and never shows up in the log as timed out. The pushbuttons itself have not set the Read flag and do show up in the logs as timed out.

Do you suggest me to check the Read flag on the pushbuttons? I’ve never unchecked that flag, so I wonder a little bit. But I see you wrote usually, so I assume defaults are not always set correctly by the manufacturer.

So grateful for your support, thanks!

farmio · July 2, 2025, 9:48am

No. You shouldn’t read a state from something that triggers an event. If a button turns on a channel of a dimming actuator, read the actuators state, not the buttons.

I’d recommend to change to group address view. In a single GA you should have at most 1 object with the read flag set.
If there is none, this address can’t be read. Which is fine for some addresses - like relative dimming etc. (stateless event triggers). In HA you shouldn’t configure those as *_state_address` though.

tominvasio · July 2, 2025, 7:44pm

Okay. I still try to figure out the problem, say, on a motion detector. Taken out one of my configuration, as example.

It has two Group Addresses,

motion (“Bewegung”) with a Write but no Read flag
brightness (“Helligkeit”), with a Read but no Write flag

Brightness is configured as sensor according to the docs as follows:

- name: "BM Zugang Helligkeit"
  state_address: "31/0/1"
  type: "illuminance"

This address is not showing up in the logs as timed out. According to your explanation, because the Read flag in ETS is set.

Motion is configured as binary sensor according to the docs as follows:

- name: "BM Zugang Bewegung"
  state_address: "31/0/0"
  device_class: "motion"

This address is showing up in the logs as timed out, as it has no Read flag set in ETS.

May I ask, what screw would you turn?

farmio · July 2, 2025, 7:50pm

Depends on how the motion sensor works. If it reports 1 when motion and 0 after some time, I’d probably set the read-flag… and test it reporting correct values when read

When it only ever sends 1, I’d not use a BinarySensor, but a telegram trigger.

Or use sync_state: False for this entity. See KNX integration documentation. But you’d have no initial value when restarting HA then (or restored value, not sure now).

tominvasio · July 12, 2025, 7:23pm

I’ve added sync_state: false to all wall pushbuttons (they send true and false when pushed and released) and motion detectors (they send true and false when motion detection is triggered and untriggered). All log entries are gone now and I’m sure thigs will work great in the near future.

Thank you so much for helping me understand the situation!
Thomas